Surprise attack bonus - Page 4

-Jrabbit · February 2, 2003, 02:59

SCG is new over there and just getting into it. Cut him some slack, rah.

rjmatsleepers · February 2, 2003, 05:17

Quote:

Originally posted by DrSpike

Hehe, um, as to your query I don't understand why we should test an increase in attack strength of one.

Your hypotheses are a little rough.....they look tricky to test to me.

I am sceptical about there being a sneak attack bonus, so I had in mind testing the hypothesis that there is a sneak attack bonus large enough to make a difference in real game play. The alternative hypothesis is that there is no sneak attack bonus (or if there is, it is too small to make a difference in game play).

The smallest sneak attack bonus for a warrior that would make a difference would increase its non-vet attack strength from 1 to 2. I suppose I really meant a bonus that increases its attack strength by one or more, but I presume the test will be pretty much the same.

As for the hypotheses being rough and tricky to test, I'm not sure what the problem is. If there is no sneak attack bonus, the expected proportion of wins in a series of warrior versus warrior contests on ground that does not have a defence bonus is .5. I assert (although I think you may disagree) that the expected proportion of wins for the attacker if there is a sneak attack bonus of 1 (or more) is about .935 (or more).

The random variable is the proportion of wins for the attacker in a series of trials. Surely it is fairly straight forward to test whether this comes from a distribution with a mean of .5 or from a distribution with a mean of .935 (or more).

(RJM is not sure whether the SDI defence will work twice, but keeps his fingers crossed.)

RJM at Sleepers

DrSpike · February 2, 2003, 06:51

OK a few things.

First, and most critical, the bonus need not increase the strength to the next integer value, in this case 2. This would be a 100% bonus anyway!

Secondly, Warrior v Warrior is not quite 50%, the defending unit gets a slight boost due to the way combat is resolved. This is actually very important for my test.....the properties of the test are not so good with errors in the expected probabilities under the null hypothesis.

Lastly you still seem to think the proportion of wins in 50 is binomially distributed. It most definitely is not. I think that is why you came up with 0.935 in your calculations.

Marquis de Sodaq · February 3, 2003, 00:02

Quote:

Originally posted by SCG
You know - Maquis de Sodaq never did run for trade advisor

No need, the trade advisor always bent to my will!

Interesting thread! rah, you're correct that nobody ever tested in multiplayer - all testing I mention in the combat thread was single player. There were a few small game details that multiplayer regulars would swear played differently than in single player. Such suspicions probably arise because there is something real behind them. I'm curious to read about what you discover in your test.

Look at all the cool smilies that have been added since I last visited.

War4ever · February 3, 2003, 02:34

let me know the results Rah...sorry i missed you guys

Scouse Gits · February 3, 2003, 03:45

Welcome back M le Marquis.

SG[1]

rjmatsleepers · February 3, 2003, 04:37

Quote:

Originally posted by DrSpike
OK a few things.

First, and most critical, the bonus need not increase the strength to the next integer value, in this case 2. This would be a 100% bonus anyway!

Secondly, Warrior v Warrior is not quite 50%, the defending unit gets a slight boost due to the way combat is resolved. This is actually very important for my test.....the properties of the test are not so good with errors in the expected probabilities under the null hypothesis.

Lastly you still seem to think the proportion of wins in 50 is binomially distributed. It most definitely is not. I think that is why you came up with 0.935 in your calculations.

I'll have to think about your second and third point, but as to the first ...

I was under the impression that the program uses integer arithmetic. Certainly the tests I have done suggest that the 50% veteran bonus for a warrior is rounded up to give it an attack strength of 2. Similarly the bonus for a veteran archer appears to round up giving it an attack strength of 5.

RJM at Sleepers

DrSpike · February 3, 2003, 05:39

Well the combat guru himself just popped in so perhaps he can convince you.

Scouse Gits · February 3, 2003, 07:15

The received wisdom is that the arithmetic is indeed integer, but based upon eighths.
IIRC a warrior warrior combat is worked out at 7 v 9 which is the advantage referred to in the Docs second point.
Don't quote me - I'm working from memory - it's all in the GL-Combat thread.

SG[1]

rah · February 3, 2003, 09:58

I tried three times this weekend to do the test. But everytime I tried there were enough players available for a game. As always, games take priority

I will try again.

But for the record, during the games I played there were 8 sneak attacks that I was involved in. 5 where I was the aggressor and 3 the defender. The attacking unit won ALL 8 combats. I'm confident that the suprise bonus will be proven soon.

4 out of 5 of my attacks where done with non-vet warriors. 3 of those 4 were attacking units on terrain with a def bonus. It's results like this that we see week after week after week that have convinced us that a bonus does exits.

RAH

-Jrabbit · February 3, 2003, 13:10

You and me tonite, rah?

Let's

rah · February 3, 2003, 13:13

As long as my daughter isn't home hogging the machine with her TEEN CHAT. arrgggg.

I'll keep an eye open for you if I can get on.

Rich

DrSpike · February 3, 2003, 13:34

Daughter: I am going to chat this evening, as long as dad isn't hogging the computer with his stupid civ game.....argghhhh.

rah · February 3, 2003, 14:22

Got that straight. Which is one reason why I play from Ming's on Friday nights. I only get those ugly looks from her on saturday nights. She can be out all evening and then return home around midnight and say, "ARE you still playing that dumb game?" hahahahhahaha

RAH

-Jrabbit · February 3, 2003, 18:58

So you have a big-screen video setup but only one comp -- AND a 15-year-old???

I'm guessing it's just a matter of time...

rah · February 3, 2003, 19:02

No we have three Comps. Two desktops and one laptop. BUT only one connection to the Internet. While we've considered hooking up the one in her bedroom, I'm not sure whether I want her to have one in there. At least in the family room, we can control the usage a bit. (what little control you actually have over teenagers

)

-Jrabbit · February 3, 2003, 20:22

Understood and agreed. The one in my teen's room does not get Internet access either.

But unless it's homework, I get first dibs.

War4ever · February 3, 2003, 21:28

and that is why i don't have kids yet, not willing to give up my internet

Alexander's Horse · February 3, 2003, 23:27

You don't have to give up your internet to have kids - just lock them out of the house.

Works for me.

Six Thousand Year Old Man · February 4, 2003, 00:15

Quote:

Originally posted by War4ever
and that is why i don't have kids yet, not willing to give up my internet

That sounds better than what I usually say.

rjmatsleepers · February 4, 2003, 05:29

Quote:

Originally posted by Scouse Gits
Don't quote me - I'm working from memory - it's all in the GL-Combat thread.

SG[1]

I've just read the GL-Combat thread so I can see where I was going wrong. Serves me right for reading (and beliveing) the manual.

(Makes mental note in future only to read the manual when all else fails.)

RJM at Sleepers

Scouse Gits · February 4, 2003, 05:58

that's why you rarely see RTFM on these threads

SG[1]

DrSpike · February 4, 2003, 08:48

Quote:

Originally posted by rjmatsleepers

I've just read the GL-Combat thread so I can see where I was going wrong. Serves me right for reading (and beliveing) the manual.

(Makes mental note in future only to read the manual when all else fails.)

RJM at Sleepers

Hehe, I lost my manual 2 weeks after buying the game.......it worked out in my favour I think.

Scouse Gits · February 4, 2003, 09:05

[OT] RJM what's Sleepers? [/OT]

Glad you found the GL of some use

SG[1]

rah · February 4, 2003, 10:49

Sorry JR, the daughter was queen last night. *sigh*

War4ever · February 4, 2003, 12:46

uh Spke, not sure if its helped your game ....

rah · February 4, 2003, 15:12

Elephant · February 4, 2003, 15:37

Quote:

Originally posted by rah
No we have three Comps. Two desktops and one laptop. BUT only one connection to the Internet. While we've considered hooking up the one in her bedroom, I'm not sure whether I want her to have one in there. At least in the family room, we can control the usage a bit. (what little control you actually have over teenagers

)

Get one of those Cable/DSL routers. The SMC one comes with wireless, a 4-port hub and can do NAT so multiple users share one line. I've got two laptops doing wireless anywhere in the house and the desktop for Real Work...

Marquis de Sodaq · February 5, 2003, 01:46

Quote:

Originally posted by Scouse Gits
Welcome back M le Marquis.

SG[1]

Thanks, good day to you, Sir Git.

rjmatsleepers, did you catch the bit about how the game uses integers, but in multiples of 8? That's the key - also, hit points are x10, giving you an order of magnitude more rounds for each battle.

Warrior vs warrior (both green) gives the attacker ~29% chance of success. Rah might be on to something, from the look of his initial runs...

rjmatsleepers · February 5, 2003, 04:17

Quote:

Originally posted by Scouse Gits
[OT] RJM what's Sleepers? [/OT]

Glad you found the GL of some use

SG[1]

Sleeper's is the name of a house.

Quote:

Originally posted by Marquis de Sodaq

rjmatsleepers, did you catch the bit about how the game uses integers, but in multiples of 8? That's the key

I caught that and read the GL entry. I hadn't realised that the win rate was so low. I rarely attack unless I think my chances of winning are close to 90%

RJM at Sleepers